Anborela
ANBORELA EUROPEAN INDEPENDENT AI
PRE-DEPLOYMENT ETHICS REVIEW GDPR COMPLIANT EU AI ACT ALIGNED 100% EUROPEAN INFRASTRUCTURE
AI System · Architecture & Design

Conversational AI built for real recovery. Outside Big Tech.

The ANBORELA system is a research-grade conversational AI designed to offer non-clinical, adaptive dialogue support for people in behavioural recovery — across substances and behavioural addictions. Built on sovereign European infrastructure, with no dependency on AWS, Azure or Google Cloud.

Non-clinical scope GDPR by design EU AI Act aligned Pre-deployment ethics review Anonymous web access
ANBORELA Assistant Research prototype
ANBORELA
This is a safe, anonymous space. I'm here to listen and support — not to judge. How are you feeling today?
User
I've been struggling lately and didn't know where to turn.
ANBORELA
Thank you for being here. That takes courage. We can talk at whatever pace feels right for you — there is no script, no judgement.
This system does not replace professional care. If you are in crisis, contact emergency services.
Design principles · Three pillars

What the system was built around

Every architectural decision derives from three non-negotiable commitments: the person in recovery, the integrity of their data, and independence from external technology ecosystems.

Patient-centred design

The system adapts its dialogue to the person's recovery stage — not a fixed protocol. Built from 30+ years of lived experience inside Spain's UCA clinical units.

  • Adaptive responses per recovery phase
  • No judgement, no clinical diagnosis
  • Available 24/7 via anonymous web link — zero friction
  • Emergency referrals hardcoded into safety layer
  • Supports substances and behavioural addictions

Data sovereignty

No personal identifiers, no medical records, no data sales. All interaction data is anonymised at point of collection and stored on encrypted European servers.

  • Anonymisation before storage — by design
  • Encrypted EU-jurisdiction servers only
  • Zero third-party data sharing
  • Full GDPR compliance from architecture layer
  • Participant withdrawal at any moment

Big Tech independence

ANBORELA does not run on AWS, Azure or GCP. The conversational model is owned and controlled by Anborela OÜ — not licensed from OpenAI, Anthropic or Meta.

  • 100% European infrastructure
  • Open to institutional architecture audit
  • EU AI Act — high-risk AI alignment
  • Registered in Estonia · EU eResidency
  • Sovereign model, not a wrapper
System architecture · How it works

How the system works

A text-based conversational model trained on public linguistic corpora, academic sources, manually curated dialogue examples and synthetic training conversations. No personal medical records used at any stage.

Step 01
User input via anonymous web link
Anonymous message. No account creation. No personal data linked to identity.
Step 02
Context analysis
Recovery stage detection. Emotional state parsing. Non-clinical framing.
Step 03
Safety layer check
Hardcoded rules: crisis detection, emergency referrals, scope boundaries.
Step 04
Adaptive response
Personalised, non-judgemental dialogue aligned to recovery phase.
Step 05
Anonymised logging
Interaction data stripped of identifiers. Encrypted EU storage for research only.
System type Text-based conversational AI Non-clinical support scope. Does not diagnose or prescribe.
Access channel Anonymous web link No account, no app, no installation. Available 24/7. Fully anonymous.
Infrastructure European · Sovereign Not hosted on AWS, Azure or GCP. EU-jurisdiction data at all times.
Current phase Pre-deployment Under voluntary ethics review. No public access until approval.
Safety & Data · Design principles

Ethics and safety by design

Designed for oversight from the first line of code. Every component is auditable, transparent and built to meet European regulatory standards before deployment.

System compliance · Research phase

Regulatory alignment

GDPR compliance 100%
EU AI Act alignment Architecture phase
Pre-deployment ethics review Submitted
Big Tech infrastructure dependency 0%
Personal data collected from users None
GDPR & data protection

What we collect and how

Anonymised at collection. No real names, IDs or medical records ever requested or stored.
EU-only storage. Encrypted European servers. Data never leaves EU jurisdiction.
Zero data sales. No transfer to third parties without explicit written consent.
Research purpose only. Interaction data used exclusively for system improvement and academic research.
Right to withdraw. Participants may exit at any time with immediate cessation of data collection.
Informed consent. Explicit consent mechanism before any interaction begins.
Feature Big Tech AI systems ANBORELA
EU-sovereign infrastructure ✕ ✓
Zero personal data collection ✕ ✓
Pre-deployment ethics review ✕ ✓
Open to institutional audit ✕ ✓
Recovery-specific adaptive dialogue ✕ ✓
Built from lived patient experience ✕ ✓

Architecture documentation available on request · ai@anborela.ee · Anborela OÜ · Estonia

Ready to co-design the evidence?

Invite your institution into the research.

We are inviting European universities, public health institutes and research bodies to join the scientific steering group. Ethics-first, GDPR-compliant, Big-Tech-independent. Open to EU and UK partners.

Start the conversation → Read the research proposal